An aerial view of lower Manhattan in 1928. (AP Photo)

"White flight" is usually described as a post-World War II phenomenon, one that required highways and suburbs and big lawns to flee to.

But whites in northern cities really began re-sorting themselves — specifically away from blacks — in the first decades of the 20th century, and what happened then remains relevant to American cities that are still racially divided today.

"If you want to understand the origins of segregation in the U.S., you have to look at this period between 1900 and 1930," says Allison Shertzer, an economist at the University of Pittsburgh who has studied detailed, digitized census forms from that era with Pittsburgh colleague Randall P. Walsh.

In their new research, they studied how the arrival of blacks in 10 northern cities at the time influenced white behavior. Over the course of the first three decades after the turn of the century, coinciding with the start of the Great Migration of blacks out of the South, this pattern accelerated: As blacks arrived in northern neighborhoods, more whites left. By the 1920s, there were more than three white departures for every black arrival.

Shertzer and Walsh, who tried to account for other reasons why neighborhood populations shifted, believe this was causal. "Whites left the neighborhood as a result of blacks arriving," Shertzer says, "not for other reasons."

The suburbs we know today effectively didn't exist at the time, so whites were leaving these neighborhoods for other neighborhoods in the city. That makes this earlier form of white flight even more striking; their new homes didn't necessarily have lower taxes or better school districts, factors that complicated the motivations of later generations of whites.

The accumulation of all those individual decisions is an important part of explaining why segregation took root in places like Baltimore, Philadelphia and Chicago, as this graph from the paper illustrates. The dissimilarity and isolation indexes are two tools researchers use to measure segregation; one captures the degree to which blacks and whites would need to move around a city to turn homogeneous neighborhoods into diverse ones, and the other captures how isolated blacks are from whites in the neighborhoods where they live.

The argument Shertzer and Walsh make is novel not just because it turns our attention to an earlier period in American history (a time from which digitized census data only recently became available). The forces they're pointing to are not the usual culprits.

"The narrative about the emergence of segregation in the U.S. has been all about these discriminatory institutions," Shertzer says, "things like violence, fire bombing, throwing bricks through people's windows, racially restrictive covenants, barriers that kept blacks out."

It's widely documented that banks in this era refused loans to redlined blacks, that real estate agents steered them away from white neighborhoods, that legal agreements ensured all-white neighborhoods stayed that way. Shertzer and Walsh are pointing to another set of factors — not the policies of institutions, but the behavior of individuals.

"We argue if you took the current legal and institutional environment and plunked it on the early 20th century," Shertzer says, "we would still have very segregated cities, because a certain number whites were unwilling to live with blacks."

Even if the Fair Housing Act had existed back then, if restrictive covenants were illegal in 1920s America, we'd have gotten segregated cities anyway because of behavior that's beyond the reach of regulation. You can create a lot segregation, this research says, without having any discriminatory institutions. Uncoordinated market choices create it.

And this is the part that's particularly relevant today. All of those historic institutional barriers have largely (but not entirely) been eliminated. But we know from research on the behavior of people searching for homes that racial preferences still shape where people choose to live today.

This is how you get a city like Ferguson, Shertzer says, with its predominantly white police force and predominantly black population. That odd pattern reflects more recent waves of white flight out of that struggling St. Louis suburb.

"When white homeowners think about how their choices shape the racial composition of neighborhoods, they say 'I would never support discrimination in mortgage lending, I'd never support anything like redlining or restrictive covenants, I would never participate in something like that,'" Shertzer says. "But the choices of white homeowners to move to neighborhoods that are almost exclusively white — this is a quantitatively important mechanism that keeps cities segregated."

The choices whites make today, though, are now also influenced by a century of segregation and the ways it has conflated race with poverty, crime and school quality. That makes this challenge a lot harder to resolve than if we had tried a century ago.